plot multiple dependence data in one graph in Julia - julia
I'm trying to plot on Julia, here is the sample data Ecdat Airline. I want to compare the airline 1~6's cost, but the x-axis was wrong(should be 1970-1984), and my code seems too dirty, maybe it could be plot in one line.
using Plots, UnicodePlots, RDatasets
unicodeplots()
v = dataset("Ecdat","Airline")
plot(v.Cost[v.Airline .== 1])
plot!(v.Cost[v.Airline .== 2])
plot!(v.Cost[v.Airline .== 3])
plot!(v.Cost[v.Airline .== 4])
plot!(v.Cost[v.Airline .== 5])
plot!(v.Cost[v.Airline .== 6])
The output
┌────────────────────────────────────────────────────────────┐
4.88870026e6 │⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⠄⠀│ y1
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡠⠔⠁⠀⠀│ y2
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⠤⠊⠀⠀⣀⠄⠀│ y3
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⣀⠤⠔⠊⠁⣀⠤⠒⠉⠀⠀⠀│ y4
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡠⠒⠒⠊⠉⠉⠀⢀⣀⠤⠒⠉⠀⠀⠀⠀⠀⠀⠀│ y5
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡠⠊⠀⠀⠀⠀⡠⠒⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│ y6
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⠔⠁⠀⠀⢀⠔⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡰⠁⠀⠀⢀⠎⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⡠⠊⠀⠀⢀⠔⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⠤⠊⠁⠀⠀⡠⠊⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⡠⠔⠊⠁⠀⠀⠀⡠⠊⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⠤⠒⠁⠀⠀⠀⠀⠀⡠⠔⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⠤⠤⠒⠊⠁⠀⠀⠀⠀⠀⣀⡠⠔⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
│⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⣀⠤⠔⠒⠉⠁⠀⠀⠀⠀⠀⢀⣀⠤⠔⠊⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⡀⠀│
│⠀⠀⠀⢀⣀⣀⡠⠤⠤⠒⠊⠉⠀⠀⠀⠀⠀⣀⡠⠤⠒⠊⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡠⠤⣤⣒⣒⣉⣁⡀⠀│
│⠀⠈⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⢀⡠⠔⠒⠉⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡠⠤⠔⣒⡲⠮⠛⠛⠉⠓⠊⠉⠉⠀⠀⠀⢀⣀⠄⠀│
│⠀⠀⠀⠀⠀⠀⢀⣀⡠⠤⠒⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⡠⠤⠔⠒⠊⠉⣁⡠⠔⠒⠉⠀⠀⠀⠀⠀⠀⠀⠀⣀⡠⠤⠒⠊⠁⠀⠀⠀│
│⠀⠐⠒⠒⠉⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣀⣀⣀⠤⠤⠤⠤⠒⠒⣒⣊⣉⣉⣁⠤⠤⠤⠔⠒⠊⠉⠀⠀⠀⠀⣀⣀⣠⡤⠤⠴⠶⠮⠭⠔⠒⠒⠒⠒⠒⠒⠂⠀│
│⠀⠠⠤⠤⠔⠒⣒⣒⣒⡲⠶⠶⠮⠛⠛⠛⠓⠒⠒⠊⠉⠉⠉⠉⠉⣉⣀⣀⣀⣀⣀⣀⣤⣤⣤⣔⡲⠶⠮⠛⠛⠋⠉⠁⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀│
-71402.26000000001 │⠤⠼⠿⠿⠿⠿⠶⠶⠶⠶⠶⠶⠶⠾⠿⠿⠿⠭⠭⠭⠭⠭⠭⠭⠭⠭⠭⠭⠭⠭⠭⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤⠤│
└────────────────────────────────────────────────────────────┘
0.5800000000000001 15.42
To plot DataFrames, use StatsPlots rather than Plots (it's an extension of Plots to DataFrames and statistical types). Also you don't need to import UnicodePlots directly. Finally, you're looking for the group keyword.
Your code should be
using StatsPlots, RDatasets
unicodeplots()
v = dataset("Ecdat","Airline")
#df v plot(:Year, :Cost, group = :Airline)
Related
How to create a boxplot in R with different characters to identify moderate and extreme outliers?
I would like to create a boxplot in R with moderate (o; <= Q1-1.5IQR; >= Q3+1.5IQR) and extreme/severe (*; <= Q1-3IQR; >= Q3+3IQR) outliers identified differently. Like this:
One way is to change the range argument and add another boxplot on top of the first. But it's difficult to suppress specific outliers, so instead of a * for the extreme outliers, I've filled in the default O (pch=1) with pch=20. set.seed(1) x <- rt(200, df=3) boxplot(x, outcol="red") boxplot(x, add=TRUE, range=3, pars=list(outpch=20, outcol="red", whisklty=0, staplelty=0)) If you really wanted asterisks, you can make them much larger so that they totally obscure the previous circles (moderate outliers). boxplot(x, outcol="red") boxplot(x, add=TRUE, range=3, pars=list(outpch='*', outcol="red", outcex=3, whisklty=0, staplelty=0))
I achieved an elegant solution:
y <- c(-4,3,3,3,4,4,5,6,7,8,12,16,26)
par(bty = 'n',mar = c(0,2,0,0))
bp <-boxplot(x = y, plot=FALSE,ylab="x",mgp = c(3, 3, 0),axes=FALSE)
#ii) Plot the boxplot, suppressing points with outpch=NA.
bxp(bp, outpch=NA)
bxp.extreme<-function(z, extreme=3.5) {
boxrange <- bp$stats[4,bp$group] - bp$stats[2,bp$group]
big.outlier<- (z$out >= bp$stats[4,bp$group] + extreme*boxrange) |
(z$out <= bp$stats[2,bp$group] - extreme*boxrange)
return(big.outlier)
}
ext<-bxp.extreme(bp)
points(bp$group, bp$out, pch=ifelse(ext, 8,1), col=ifelse(ext, 1,1))
legend(title = "Outliers","topright", bty="0", legend=c(expression(atop("Q"[1]*"-1.5IQR ≤ Moderate ≤ Q"[3]*"-3IQR","Q"[3]*"+1.5IQR ≤ Moderate ≤ Q"[3]*"+3IQR")),
expression(atop("Extreme ≤ Q"[1]*"-3IQR","Extreme ≥ Q"[3]*"+3IQR"))),
pch=c(1,8), col=c(1,1),y.intersp=1.5)
I used the code found here and adapted it.
Octave one line two plot differ from multiple line plot
I'm fitting some data, but when I plot the data and fit in one-line-plot (see left figure) the drawn graph is correct plot(x, y, '.b;data;', [0.05 2], phi(1)+phi(2)*[0.05 2], '--r;fit;', [0.05 2]); But when I used two separated plot (see right figure) the graph differs from mention above hold on; plot(x, y, '.b;data;', [0.05 2]); plot(phi(1)+phi(2)*[0.05 2], '--r;fit;', [0.05 2]); hold off; grid on; Data [x, y] ans = 0.050000 3571.000000 0.100000 6567.000000 0.200000 12760.000000 0.300000 20512.000000 0.400000 25480.000000 0.500000 32088.000000 1.000000 63223.000000 2.000000 128690.000000 Calculate Linear Regression A = [N, sum(x); sum(x), sum(x.*x)]; b = [sum(y); sum(x.*y)]; phi = inv(A)*b; Is there any way to solve this?
When the x-axis values are not specified then 1:numel(y) are considered to be the x-axis values. In your code: %Your first graph: plot(x, y, '.b;data;', [0.05 2], phi(1)+phi(2)*[0.05 2], '--r;fit;', [0.05 2]); %^^^^^^^ %Your second graph: plot(x, y, '.b;data;', [0.05 2]); %^^^^^^^^^ plot(phi(1)+phi(2)*[0.05 2], '--r;fit;', [0.05 2]); %!!!!!!!!!!!!!!!!!!!!!! %^^^^^^^^^ The parts highlighted with ^ in above plot commands have x=[1 2] and y=[0.05 2] and the part highlighted with ! has x=[1 2] and y=phi(1)+phi(2)*[0.05 2]. You can see those lines being drawn if you zoom at that area. So your first plot command should be: plot(x, y, '.b;data;', [0.05 2], phi(1)+phi(2)*[0.05 2], '--r;fit;'); and it should be split like this: plot(x, y, '.b;data;'); hold on; plot([0.05 2], phi(1)+phi(2)*[0.05 2], '--r;fit;');
R Plot - Y Axis and plotted values are not the same (values from txt file)
I am somewhat new to R and I am trying to plot values from real data . x axis is the time and y axis real values but in graph y axis is between 0 and 100, but the real values of y are different. I need your help.Thank you
y <- c(9.21368, 8.90582, 8.92218, 8.53934, 7.99324, 7.53952, 7.51255, 7.1065, 6.91431, 6.9704, 7.92692, 7.74947, 7.10618, 6.50479, 5.9741, 5.08136, 4.32405, 3.86809, 3.4377, 3.54596, 3.07406, 2.92242, 1.67285, 1.11221, 0.501673) t <- 1 : length(y) plot(t, y) # Length(y) and length(t) must be same.
You can view your y-axis values in the plot, by adding text function y <- c(9.21368, 8.90582, 8.92218, 8.53934, 7.99324, 7.53952, 7.51255, 7.1065, 6.91431, 6.9704, 7.92692, 7.74947, 7.10618, 6.50479, 5.9741, 5.08136, 4.32405, 3.86809, 3.4377, 3.54596, 3.07406, 2.92242, 1.67285, 1.11221, 0.501673) t <- 1: length(y) plot(t,y) text(t,y,pos=3, cex=0.7)
The y-axis is always like the x axis but in reality is not like that:
3D with value interpolation in R (X, Y, Z, V)
Is there an R package that does X, Y, Z, V interpolation? I see that Akima does X, Y, V but I need one more dimension. Basically I have X,Y,Z coordinates plus the value (V) that I want to interpolate. This is all GIS data but my GIS does not do voxel interpolation So if I have a point cloud of XYZ coordinates with a value of V, how can I interpolate what V would be at XYZ coordinate (15,15,-12) ? Some test data would look like this: X <-rbind(10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,10,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,20,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,30,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,40,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50,50) Y <- rbind(10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50,10,10,10,10,10,20,20,20,20,20,30,30,30,30,30,40,40,40,40,40,50,50,50,50,50) Z <- rbind(-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-5,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-17,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29,-29) V <- rbind(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,5,25,35,75,25,50,0,0,0,0,0,10,12,17,22,27,32,37,25,13,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,50,125,130,105,110,115,165,180,120,100,80,60,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)
I had the same question and was hoping for an answer in R.
My question was: How do I perform 3D (trilinear) interpolation using regular gridded coordinate/value data (x,y,z,v)? For example, CT images, where each image has pixel centers (x, y) and greyscale value (v) and there are multiple image "slices" (z) along the thing being imaged (e.g., head, torso, leg, ...).
There is a slight problem with the given example data.
# original example data (reformatted)
X <- rep( rep( seq(10, 50, by=10), each=25), 3)
Y <- rep( rep( seq(10, 50, by=10), each=5), 15)
Z <- rep(c(-5, -17, -29), each=125)
V <- rbind(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,5,25,35,75,25,50,0,0,0,0,0,10,12,17,22,27,32,37,25,13,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,50,125,130,105,110,115,165,180,120,100,80,60,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)
# the dimensions of the 3D grid described do not match the number of values
(length(unique(X))*length(unique(Y))*length(unique(Z))) == length(V)
## [1] FALSE
## which makes sense since 75 != 375
# visualize this:
library(rgl)
plot3d(x=X, y=Y, z=Z, col=terrain.colors(181)[V])
# examine the example data real quick...
df <- data.frame(x=X,y=Y,z=Z,v=V);
head(df);
table(df$x, df$y, df$z);
# there are 5 V values at each X,Y,Z coordinate... duplicates!
# redefine Z so there are 15 unique values
# making 375 unique coordinate points
# and matching the length of the given value vector, V
df$z <- seq(-5, -29, length.out=15)
head(df)
table(df$x, df$y, df$z);
# there is now 1 V value at each X,Y,Z coordinate
# that was for testing, now actually redefine the Z vector.
Z <- rep(seq(-5,-29, length.out = 15), 25)
# plot it.
library(rgl)
plot3d(x=X, y=Y, z=Z, col=terrain.colors(181)[V])
I couldn't find any 4D interpolation functions in the usual R packages, so I wrote a quick and dirty one. The following implements (without ANY error checking... caveat emptor!) the technique described at: https://en.wikipedia.org/wiki/Trilinear_interpolation
# convenience function #1:
# define a function that takes a vector of lookup values and a value to lookup
# and returns the two lookup values that the value falls between
between = function(vec, value) {
# extract list of unique lookup values
u = unique(vec)
# difference vector
dvec = u - value
vals = c(u[dvec==max(dvec[dvec<0])], u[dvec==min(dvec[dvec>0])])
return(vals)
}
# convenience function #2:
# return the value (v) from a grid data.frame for given point (x, y, z)
get_value = function(df, xi, yi, zi) {
# assumes df is data.frame with column names: x, y, z, v
subset(df, x==xi & y==yi & z==zi)$v
}
# inputs df (x,y,z,v), points to look up (x, y, z)
interp3 = function(dfin, xin, yin, zin) {
# TODO: check if all(xin, yin, zin) equals a grid point, if so just return the point value
# TODO: check if any(xin, yin, zin) equals a grid point, if so then do bilinear or linear interp
cube_x <- between(dfin$x, xin)
cube_y <- between(dfin$y, yin)
cube_z <- between(dfin$z, zin)
# find the two values in each dimension that the lookup value falls within
# and extract the cube of 8 points
tmp <- subset(dfin, x %in% cube_x &
y %in% cube_y &
z %in% cube_z)
stopifnot(nrow(tmp)==8)
# define points in a periodic and cubic lattice
x0 = min(cube_x); x1 = max(cube_x);
y0 = min(cube_y); y1 = max(cube_y);
z0 = min(cube_z); z1 = max(cube_z);
# define differences in each dimension
xd = (xin-x0)/(x1-x0); # 0.5
yd = (yin-y0)/(y1-y0); # 0.5
zd = (zin-z0)/(z1-z0); # 0.9166666
# interpolate along x:
v00 = get_value(tmp, x0, y0, z0)*(1-xd) + get_value(tmp,x1,y0,z0)*xd # 2.5
v01 = get_value(tmp, x0, y0, z1)*(1-xd) + get_value(tmp,x1,y0,z1)*xd # 0
v10 = get_value(tmp, x0, y1, z0)*(1-xd) + get_value(tmp,x1,y1,z0)*xd # 0
v11 = get_value(tmp, x0, y1, z1)*(1-xd) + get_value(tmp,x1,y1,z1)*xd # 65
# interpolate along y:
v0 = v00*(1-yd) + v10*yd # 1.25
v1 = v01*(1-yd) + v11*yd # 32.5
# interpolate along z:
return(v0*(1-zd) + v1*zd) # 29.89583 (~91.7% between v0 and v1)
}
> interp3(df, 15, 15, -12)
[1] 29.89583
Testing that same source's assertion that trilinear is simply linear(bilinear(), bilinear()), we can use the base R linear interpolation function, approx(), and the akima package's bilinear interpolation function, interp(), as follows:
library(akima)
approx(x=c(-11.857143,-13.571429),
y=c(interp(x=df[round(df$z,1)==-11.9,"x"], y=df[round(df$z,1)==-11.9,"y"], z=df[round(df$z,1)==-11.9,"v"], xo=15, yo=15)$z,
interp(x=df[round(df$z,1)==-13.6,"x"], y=df[round(df$z,1)==-13.6,"y"], z=df[round(df$z,1)==-13.6,"v"], xo=15, yo=15)$z),
xout=-12)$y
# [1] 0.2083331
Checked another package to triangulate:
library(oce)
Vmat <- array(data = V, dim = c(length(unique(X)), length(unique(Y)), length(unique(Z))))
approx3d(x=unique(X), y=unique(Y), z=unique(Z), f=Vmat, xout=15, yout=15, zout=-12)
[1] 1.666667
So 'oce', 'akima' and my function all give pretty different answers. This is either a mistake in my code somewhere, or due to differences in the underlying Fortran code in the akima interp(), and whatever is in the oce 'approx3d' function that we'll leave for another day.
Not sure what the correct answer is because the MWE is not exactly "minimum" or simple. But I tested the functions with some really simple grids and it seems to give 'correct' answers. Here's one simple 2x2x2 example:
# really, really simple example:
# answer is always the z-coordinate value
sdf <- expand.grid(x=seq(0,1),y=seq(0,1),z=seq(0,1))
sdf$v <- rep(seq(0,1), each=4)
> interp3(sdf,0.25,0.25,.99)
[1] 0.99
> interp3(sdf,0.25,0.25,.4)
[1] 0.4
Trying akima on the simple example, we get the same answer (phew!):
library(akima)
approx(x=unique(sdf$z),
y=c(interp(x=sdf[sdf$z==0,"x"], y=sdf[sdf$z==0,"y"], z=sdf[sdf$z==0,"v"], xo=.25, yo=.25)$z,
interp(x=sdf[sdf$z==1,"x"], y=sdf[sdf$z==1,"y"], z=sdf[sdf$z==1,"v"], xo=.25, yo=.25)$z),
xout=.4)$y
# [1] 0.4
The new example data in the OP's own, accepted answer was not possible to interpolate with my simple interp3() function above because:
(a) the grid coordinates are not regularly spaced, and
(b) the coordinates to lookup (x1, y1, z1) lie outside of the grid.
# for completeness, here's the attempt:
options(scipen = 999)
XCoor=c(78121.6235,78121.6235,78121.6235,78121.6235,78136.723,78136.723,78136.723,78136.8969,78136.8969,78136.8969,78137.4595,78137.4595,78137.4595,78125.061,78125.061,78125.061,78092.4696,78092.4696,78092.4696,78092.7683,78092.7683,78092.7683,78092.7683,78075.1171,78075.1171,78064.7462,78064.7462,78064.7462,78052.771,78052.771,78052.771,78032.1179,78032.1179,78032.1179)
YCoor=c(5213642.173,523642.173,523642.173,523642.173,523594.495,523594.495,523594.495,523547.475,523547.475,523547.475,523503.462,523503.462,523503.462,523426.33,523426.33,523426.33,523656.953,523656.953,523656.953,523607.157,523607.157,523607.157,523607.157,523514.671,523514.671,523656.81,523656.81,523656.81,523585.232,523585.232,523585.232,523657.091,523657.091,523657.091)
ZCoor=c(-3.0,-5.0,-10.0,-13.0,-3.5,-6.5,-10.5,-3.5,-6.5,-9.5,-3.5,-5.5,-10.5,-3.5,-5.5,-7.5,-3.5,-6.5,-11.5,-3.0,-5.0,-9.0,-12.0,-6.5,-10.5,-2.5,-3.5,-8.0,-3.5,-6.5,-9.5,-2.5,-6.5,-8.5)
V=c(2.4000,30.0,620.0,590.0,61.0,480.0,0.3700,0.0,0.3800,0.1600,0.1600,0.9000,0.4100,0.0,0.0,0.0061,6.0,52.0,0.3400,33.0,235.0,350.0,9300.0,31.0,2100.0,0.0,0.0,10.5000,3.8000,0.9000,310.0,0.2800,8.3000,18.0)
adf = data.frame(x=XCoor, y=YCoor, z=ZCoor, v=V)
# the first y value looks like a typo?
> head(adf)
x y z v
1 78121.62 5213642.2 -3.0 2.4
2 78121.62 523642.2 -5.0 30.0
3 78121.62 523642.2 -10.0 620.0
4 78121.62 523642.2 -13.0 590.0
5 78136.72 523594.5 -3.5 61.0
6 78136.72 523594.5 -6.5 480.0
x1=198130.000
y1=1913590.000
z1=-8
> interp3(adf, x1,y1,z1)
numeric(0)
Warning message:
In min(dvec[dvec > 0]) : no non-missing arguments to min; returning Inf
Whether the test data did or not make sense, I still needed an algorithm. Test data is just that, something to fiddle with and as a test data it was fine. I wound up programming it in python and the following code takes XYZ V and does a 3D Inverse Distance Weighted (IDW) interpolation where you can set the number of points used in the interpolation. This python recipe only interpolates to one point (x1, y1, z1) but it is easy enough to extend. import numpy as np import math #34 points XCoor=np.array([78121.6235,78121.6235,78121.6235,78121.6235,78136.723,78136.723,78136.723,78136.8969,78136.8969,78136.8969,78137.4595,78137.4595,78137.4595,78125.061,78125.061,78125.061,78092.4696,78092.4696,78092.4696,78092.7683,78092.7683,78092.7683,78092.7683,78075.1171,78075.1171,78064.7462,78064.7462,78064.7462,78052.771,78052.771,78052.771,78032.1179,78032.1179,78032.1179]) YCoor=np.array([5213642.173,523642.173,523642.173,523642.173,523594.495,523594.495,523594.495,523547.475,523547.475,523547.475,523503.462,523503.462,523503.462,523426.33,523426.33,523426.33,523656.953,523656.953,523656.953,523607.157,523607.157,523607.157,523607.157,523514.671,523514.671,523656.81,523656.81,523656.81,523585.232,523585.232,523585.232,523657.091,523657.091,523657.091]) ZCoor=np.array([-3.0,-5.0,-10.0,-13.0,-3.5,-6.5,-10.5,-3.5,-6.5,-9.5,-3.5,-5.5,-10.5,-3.5,-5.5,-7.5,-3.5,-6.5,-11.5,-3.0,-5.0,-9.0,-12.0,-6.5,-10.5,-2.5,-3.5,-8.0,-3.5,-6.5,-9.5,-2.5,-6.5,-8.5]) V=np.array([2.4000,30.0,620.0,590.0,61.0,480.0,0.3700,0.0,0.3800,0.1600,0.1600,0.9000,0.4100,0.0,0.0,0.0061,6.0,52.0,0.3400,33.0,235.0,350.0,9300.0,31.0,2100.0,0.0,0.0,10.5000,3.8000,0.9000,310.0,0.2800,8.3000,18.0]) def Distance(x1,y1,z1, Npoints): i=0 d=[] while i < 33: d.append(math.sqrt((x1-XCoor[i])*(x1-XCoor[i]) + (y1-YCoor[i])*(y1-YCoor[i]) + (z1-ZCoor[i])*(z1-ZCoor[i]) )) i = i + 1 distance=np.array(d) myIndex=distance.argsort()[:Npoints] weightedNum=0 weightedDen=0 for i in myIndex: weightedNum=weightedNum + (V[i]/(distance[i]*distance[i])) weightedDen=weightedDen + (1/(distance[i]*distance[i])) InterpValue=weightedNum/weightedDen return InterpValue x1=198130.000 y1=1913590.000 z1=-8 print(Distance(x1,y1,z1, 12))
Create a function with multiple parameters in R
I want to compute the following functions :
here, g(x) is the density function of a distribution. I want to compute this function for several distributions. In addition, I use the library fitdistrplus.
To create g, I use the function do.call this way :
g<-function(x) {do.call(paste("d",i,sep=""),c(list(x=x),fti$estimate))}
fti$estimate contains the parameters of the distribution i.
G(x) is the cumulative distribution computed this way :
G<-function(x) {do.call(paste("p",i,sep=""),c(list(q=x),fti$estimate))}
I compute f(x) this way :
f<function(n,x) {n*g(x)*(1-G(x))^(n-1)
At last, I compute h(x) this way :
h<- function(n) {integrate(function(x) {x*f(n,x)},0,Inf)}
However, I can't plot these functions, I get the following errors :
1: In n*g(x):
Longer object length is not a multiple of shorter object length
2: In (1-G(x))^(n-1):
Longer object length is not a multiple of shorter object length
3: In x*f(n,x) :
Longer object length is not a multiple of shorter object length
Beyond, if I juste want to plot f(n,x), I get this error :
Error in list(x=x) :'x' is missing
The minimal snipset I have is the following
#i can be "exp" "lnorm" "norm" etc...
for( i in functionsName) {
png(paste(fileBase,"_",i,"_","graphics.png",sep=""))
plot.new()
fti<-fitdist(data, i)
plotdist(data,i, para=as.list(fti[[1]]))
#fti is a datatable or datafram
#fti$estimate looks like this :
# meanlog sdlog
#8.475449 1.204958
#g
pdf<-function(x) {do.call(paste("d",i,sep=""), c(list(x=x),fti$estimate))}
#G
cdf<-function(x) do.call(paste("p",i,sep=""), c(list(q=x),fti$estimate))
#f
minLaw<- function(n,x) {n*pdf(x)*(1-cdf(x))^(n-1)}
#h
minExpectedValue<-function(n) {integrate(function(x) {x*minLaw(n,x)},0,Inf)}
#these 2 following lines give an error
plot(minExpectedValue)
plot(minLaw)
dev.off()
}
I had to do some reverse engineering to figure out your d1, q1 etc calls, but I think this is how you do it. Perhaps the original problem lies in a function call like f(n=2:3, x=1:9); in such a call n should be a single value, not a vector of values.
Even if length of x was a multiple of n length, the output would most likely not be what you really wanted.
If you try to give n a vector form, you might end up in a recycled (false) output:
> print(data.frame(n=2:3, x=1:6))
- n x
1 2 1
2 3 2
3 2 3
4 3 4
5 2 5
6 3 6
where x would be evaluated with n=2 at point x=1, n=3 at point x=2 etc. What you really would've wanted is something in the lines of:
> print(expand.grid(x=1:5, n=2:3))
- x n
1 1 2
2 2 2
3 3 2
4 4 2
5 5 2
6 1 3
7 2 3
8 3 3
9 4 3
10 5 3
You could do this by calling separately for each n value:
lapply(2:3, FUN=function(n) (f(n, x=1:5)))
#[[1]]
#[1] 0.0004981910 0.0006066275 0.0007328627 0.0008786344 0.0010456478
#
#[[2]]
#[1] 0.0007464956 0.0009087272 0.0010974595 0.0013152213 0.0015644676
Did you use the same fti for all the distribution fits, even though it should've been different? Or does the i in fti refer to index i and it was a list of fits in form of ft[[i]]?
Below is a wrapper function, which is called separately for each n-value (and distribution i):
wrapper <- function(i, x, n, fti){
# As was provided by OP
g<-function(x) {do.call(paste("d",i,sep=""),c(list(x=x),fti$estimate))}
G<-function(x) {do.call(paste("p",i,sep=""),c(list(q=x),fti$estimate))}
# does the i in fti refer to fit of i:th distribution, i.e. should it be a list where i:th location in ft is i:th distribution estimates?
f<-function(n,x) {n*g(x)*(1-G(x))^(n-1)}
# was missing a '-' and a '}'
h<- function(n) {integrate(function(x) {x*f(n,x)},0,Inf)}
list(gres = g(x), Gres = G(x), fres = f(n,x), hres = h(n))
}
# Example data
require("fitdistrplus")
data(groundbeef)
serving <- groundbeef$serving
# Gumbel distribution
d1 <- function(x, a, b) 1/b*exp((a-x)/b)*exp(-exp((a-x)/b))
p1 <- function(q, a, b) exp(-exp((a-q)/b))
q1 <- function(p, a, b) a-b*log(-log(p))
fti1 <- fitdist(serving, "1", start=list(a=10, b=10))
#> fti1$estimate
# a b
#56.95893 29.07871
# Normal distribution
# dnorm, pnorm and qnorm are available in the default environment
d2 <- dnorm
p2 <- pnorm
q2 <- qnorm
fti2 <- fitdist(serving, "2", start=list(mean=0, sd=1))
#> fti2$estimate
# mean sd
#73.67743 35.92581
# Sequence of x-values
xs <- seq(-100, 100, by=1)
print((resultdist1n2 <- wrapper(i=1, x=xs, n=2, fti=fti1))$hres)
print((resultdist1n3 <- wrapper(i=1, x=xs, n=3, fti=fti1))$hres)
print((resultdist2n2 <- wrapper(i=2, x=xs, n=2, fti=fti2))$hres)
print((resultdist2n3 <- wrapper(i=2, x=xs, n=3, fti=fti2))$hres)
plot(xs, resultdist1n2$fres, col=1, type="l", ylim=c(0,0.025), xlab="x", ylab="f(n, x)")
points(xs, resultdist1n3$fres, col=2, type="l")
points(xs, resultdist2n2$fres, col=3, type="l")
points(xs, resultdist2n3$fres, col=4, type="l")
legend("topleft", legend=c("Gamma (i=1) n=2", "Gamma (i=1) n=3", "Normal (i=2) n=2", "Normal (i=2) n=3"), col=1:4, lty=1)
And the results of your desired h as found in resultdist1n2$hres etc:
h(n=2) for distribution i=1:
53.59385 with absolute error < 0.00022
h(n=3) for distribution i=1:
45.23146 with absolute error < 4.5e-05
h(n=2) for distribution i=2:
53.93748 with absolute error < 1.1e-05
h(n=3) for distribution i=2:
44.06331 with absolute error < 2e-05
EDIT: Here's how one uses the lapply function to call for each of the vector of n values 0<=n<=256:
ns <- 0:256
res1 <- lapply(ns, FUN=function(nseq) wrapper(i=1, x=xs, n=nseq, fti=fti1))
par(mfrow=c(1,2))
plot.new()
plot.window(xlim=c(-100,100), ylim=c(0, 0.05))
box(); axis(1); axis(2); title(xlab="x", ylab="f(n,x)", main="f(n,x) for gamma (i=1), n=0:256")
for(i in 1:length(ns)) points(xs, res1[[i]]$fres, col=rainbow(257)[i], type="l")
# perform similarly for the other distributions by calling with i=2, fti=fti2
# h as a function of n for dist i=1
plot(ns, unlist(lapply(res1, FUN=function(x) x$hres$value)), col=rainbow(257), xlab="n", ylab="h(n)", main="h(n) for gamma (i=1), n=0:256")
I would plot each distribution i separately like this.
The problem is that the plot method for a function expects that the function will be vectorised. In other words, if given an argument of length N, it should return a vector of results also of length N. Your minExpectedValue doesn't satisfy this; it expects that n will be a scalar, and returns a scalar. You can quickly fix this up with Vectorize. You also need to specify the name of the argument to plot over, in this case n. minExpectedValue <- Vectorize(minExpectedValue) plot(minExpectedValue, xname="n")